Climate predicts the proportion of divaricate plant species in New Zealand arborescent assemblages

Aim Puzzling features of plants are sometimes explained as legacies of co‐evolution with extinct herbivores. One example is the convergent evolution of a small‐leaved, twiggy ‘divaricate’ form in > 50 woody species in New Zealand. This growth form was first interpreted as a response to the Plio‐Pleistocene onset of frosty, droughty environments, but opinion now favours the hypothesis that it arose as a defence against large herbivorous birds (moa). It has been argued that the extinction of moa during the last millennium has left divaricate plants at risk of being out‐competed by faster growing broadleaved competitors, yet their current abundance in some habitats suggests this growth form might confer other advantages besides protection against avian browsing. We aimed to clarify the adaptive significance of the divaricate growth form by identifying environmental correlates of its geographic distribution in New Zealand. Location The main islands of the New Zealand archipelago, spanning latitude c . 34.5–47° S. Methods We used generalized additive models to examine the environmental correlates of the proportion of local tree and shrub assemblages contributed by divaricate species, on 1152 0.04 ha plots distributed across a wide range of climate and landforms. Results A model with four climatic variables explained 37% of the observed variation. Mean minimum temperature of the coldest month was the single best predictor of the proportion of divaricate species in arborescent assemblages, with annual solar radiation, vapour pressure deficit and annual water deficit also contributing to the final model. Divaricates made their largest contributions to assemblages in the rain shadow of the Southern Alps, on sites with cold winters, low solar radiation and significant vapour pressure deficits, where potential evapotranspiration exceeds precipitation by > 130 mm. No physiographic variables (e.g. slope) contributed to the final model. Main conclusions The distribution of divaricate plants suggests they cope with frosty, droughty environments better than most other elements of New Zealand's arborescent flora. This may reflect selection for both (1) physiological resistance to frost and drought and (2) deterrence of herbivores, as climatic constraints on plant growth rates probably leave seedlings exposed for longer to ground‐dwelling browsers. However, the widespread persistence of divaricate plants during the c . 450‐year hiatus between the extinction of moa and the onset of widespread browsing by introduced ungulates suggests that even in the absence of large herbivores, divaricates are well adapted to frosty, droughty environments.